Water pump operation state big data analysis and prediction method and system

Abstract

The invention relates to the technical field of water pump operation state big data analysis and prediction, and particularly discloses a water pump operation state big data analysis and prediction method and system. Reference data in a healthy operation state are obtained in a water pump operation mode switching process and are matched and compared with real-time parameter data of actual operation; and obtaining difference data reflecting the deviation degree between the actual operation state and the healthy operation state of the water pump. The method can effectively solve the problems that in the prior art, transient data are difficult to accurately recognize in a quick mode switching scene of the water pump, data analysis is difficult, and the abnormal running state of the water pump cannot be effectively predicted. By analyzing the difference data, the method can judge whether parts in the water pump are abnormal or not, so that early warning of potential risks of the water pump is achieved, non-planned shutdown caused by sudden abnormity is avoided, the stability and reliability of operation of the water pump are improved, and a powerful guarantee is provided for the production process of an enterprise.

Description

Technical Field

[0001] This application relates to the field of big data analysis and prediction technology for water pump operating status, and more specifically, to a method and system for big data analysis and prediction of water pump operating status. Background Technology

[0002] In large-scale industrial production facilities, water pumps are the core equipment for fluid transportation, and their stable operation directly affects the continuity and efficiency of the entire production process. To ensure the reliability of these critical devices and to shift from reactive maintenance to proactive preventative maintenance, many companies have introduced advanced water pump operation status analysis and prediction systems. These systems typically continuously collect various operating data from the water pumps, such as vibration signals, bearing temperature, pump body pressure, and motor current, and utilize data analysis technology to identify potential equipment problems in advance, thereby effectively avoiding unplanned downtime caused by sudden anomalies.

[0003] However, in practical applications, especially when pumps need to frequently and rapidly switch between two drastically different operating modes, existing big data analysis and prediction methods and systems for pump operating status face severe challenges. For example, switching from a high-flow, high-head batch delivery mode to a low-flow, high-precision control fine-mixing mode may only last tens of seconds to several minutes. During this period, the data stream acquired by the system often contains a large number of transient signals transitioning from one stable state to another. These transient data contain mixed characteristics of the two modes; for example, during deceleration, the vibration spectrum may simultaneously contain residual frequency components at high speeds and newly excited frequency components at low speeds; pressure sensors may record instantaneous pressure shocks generated when valves close or open rapidly. Because the input data contains a large amount of mixed operating conditions and transient change information that is not clearly distinguished, existing big data analysis and prediction methods and systems for pump operating status face significant challenges in processing them. The method does not fully grasp the complex dynamic behavior during these rapid switching processes; therefore, it tends to treat these complex signals as normal systemic noise or environmental interference, rather than component anomalies that may occur under specific operating conditions. For example, when switching from a high-flow-rate mode to a low-flow-rate mode, the rapid change in fluid velocity inside the pump can cause transient cavitation, generating broadband vibration and noise. Existing systems may misinterpret this normal physical fluctuation as early signs of bearing wear or impeller damage, resulting in frequent false alarms. These false alarms may not be genuine equipment malfunctions, but rather reflections of the system's insufficient understanding of transient operating conditions.

[0004] To address the aforementioned issues, existing technologies urgently need improvement. Summary of the Invention

[0005] This application provides a big data analysis and prediction method and system for water pump operation status, which aims to solve the problem that existing technologies have difficulty in accurately identifying transient data during water pump operation mode switching, resulting in difficulties in data analysis and the inability to effectively predict abnormal water pump operation status.

[0006] To solve the above problems, the solution proposed in this application is as follows:

[0007] As one aspect of this application, a big data analysis and prediction method for water pump operating status is provided. This method is applied in a water pump control system and includes:

[0008] In water pump operation applications, obtain reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operation mode to the second water pump operation mode under healthy operating conditions;

[0009] In the same water pump operation application, when the water pump actually switches from the first water pump operation mode to the second water pump operation mode, the real-time operation parameter data of the water pump is acquired, and the real-time operation parameter data of the water pump is matched and compared with the reference data to obtain the difference data reflecting the degree of deviation between the actual operation state of the water pump and the healthy operation state of the water pump.

[0010] The analysis of the discrepancies will determine whether any components inside the water pump are malfunctioning.

[0011] Furthermore, the reference data includes one or more combinations of vibration data, temperature data, or pressure data.

[0012] Furthermore, the step of obtaining reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operating mode to the second water pump operating mode under healthy operating conditions in the water pump operation application specifically includes:

[0013] Obtain reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operating mode to the second water pump operating mode under a healthy operating scenario;

[0014] Repeat the above operation a preset number of times to obtain a data set containing reference data on the changes of various operating parameters of multiple water pumps over time;

[0015] The reference data on the changes of various operating parameters of multiple water pumps over time are accumulated and averaged to obtain the processed reference data on the changes of various operating parameters of the water pumps over time.

[0016] Furthermore, the step of analyzing the difference data to determine whether the pump's operating status is abnormal in the second pump operating mode specifically includes:

[0017] The difference data are fitted using a linear interpolation method to obtain the difference signal curve;

[0018] Identify the cumulative duration corresponding to curve segments that exceed a set difference threshold from the difference signal curve graph;

[0019] The cumulative duration is used to determine whether there is any abnormality in the components inside the water pump. When the cumulative duration exceeds a set duration threshold, it is determined that there is an abnormality in the components inside the water pump.

[0020] Furthermore, the step of determining whether there is an abnormality in a component inside the water pump based on the accumulated time, wherein when the accumulated time exceeds a set time threshold, after determining that there is an abnormality in a component inside the water pump, includes:

[0021] Obtain the basic feature distribution of the differential signal curve;

[0022] Obtain the sensitivity parameters of the fluid in the current pump from a priori database of fluid property sensitivity parameters;

[0023] Based on the sensitivity parameters and basic characteristic distribution of the fluid in the current water pump, the reference signal curve is obtained by reconstructing the hydrodynamic equation and combining iterative algorithm.

[0024] The difference signal curve is compared with the reference signal curve to verify whether the actual operating state of the water flow is abnormal due to the characteristic shift caused by the change in fluid properties.

[0025] Furthermore, after analyzing the difference data to determine whether any components inside the water pump are malfunctioning, the process includes:

[0026] When it is determined that a component inside the water pump is abnormal, the entropy value of the difference data is calculated using the entropy value method, and a timestamp is assigned to the entropy value to form structured data. This structured data is then stored in the historical entropy value database.

[0027] After each water pump actually switches from the first water pump operation mode to the second water pump operation mode, the structured data stored in the historical entropy value database is extracted according to a preset time period.

[0028] Analysis of multiple extracted structural data points revealed that when the entropy value in the extracted structural data showed a continuous increasing trend over time and the slope of this continuous increasing trend exceeded a set slope threshold, it was confirmed that the properties of the material being transported in the water pump were abnormal.

[0029] Furthermore, after analyzing the difference data to determine whether any components inside the water pump are malfunctioning, the process includes:

[0030] Extract prior knowledge about the relationship between pump flow rate and easily worn parts of the pump from the enterprise's pump operation database, and build a pump operation knowledge base.

[0031] When it is confirmed that the water pump is in an abnormal operating state, the water pump flow information of the current water pump in the second water pump operation mode is collected, and the associated water pump wear parts are matched from the water pump operation knowledge base based on the water pump flow information.

[0032] Analysis of easily worn parts of the water pump and differential signal curves is used to identify abnormal components within the water pump.

[0033] Furthermore, the step of analyzing easily worn parts of the water pump and the differential signal curve to identify abnormal components within the water pump specifically includes:

[0034] The difference points exceeding the set difference amplitude are identified from the difference signal curve graph, and the corresponding water pump operating parameter data are obtained by reverse extraction based on the difference points.

[0035] The pump operating parameter data corresponding to the difference point was obtained by reverse extraction, and the abnormal components in the pump operating parameter data corresponding to the difference point were identified by time-frequency analysis.

[0036] Based on the abnormal component, the suspected wear component causing the abnormal component is obtained. The suspected wear component is matched with the easily worn components of the water pump. Based on the intersection between the suspected wear component and the easily worn components of the water pump, the component that caused the abnormality is confirmed.

[0037] Further, after the steps of obtaining the suspected wear component based on the abnormal component, matching the suspected wear component with the easily worn components of the water pump, and confirming the abnormal component based on the intersection between the suspected wear component and the easily worn components of the water pump, the process includes:

[0038] Acquire the first operating parameter data when the water pump is in the first operating mode and acquire the second operating parameter data when the water pump is in the second operating mode;

[0039] Based on the confirmed abnormal component, first abnormality information and second abnormality information about the abnormal component are extracted from the first operating parameter data and the second operating parameter data;

[0040] By associating the first abnormality information, the second abnormality information, and the difference data, a development path indication is formed for the component that has an abnormality.

[0041] As a second aspect of this application, a big data analysis and prediction system for water pump operating status is provided, comprising:

[0042] The health parameter acquisition module is used to acquire reference data on the changes of various operating parameters of the water pump over time when the water pump is operating healthily and switching from the first water pump operating mode to the second water pump operating mode.

[0043] The real-time parameter acquisition and difference signal generation module is used to acquire real-time operating parameter data of the water pump when the water pump actually switches from the first water pump operating mode to the second water pump operating mode in the same water pump operating application. The real-time operating parameter data of the water pump is matched and compared with reference data to obtain difference data reflecting the degree of deviation between the actual operating state of the water pump and the healthy operating state of the water pump.

[0044] The data analysis module is used to analyze the difference data to determine whether there are any abnormalities in the components inside the water pump.

[0045] As described above, the big data analysis and prediction method for water pump operating status provided in this application acquires reference data under healthy operating conditions during water pump operating mode switching and compares it with real-time parameter data of actual operation to obtain difference data reflecting the degree of deviation between the actual operating status and the healthy operating status of the water pump. This method effectively solves the problem in existing technologies where it is difficult to accurately identify transient data in water pump rapid mode switching scenarios, leading to difficulties in data analysis and an inability to effectively predict abnormal water pump operating status. By analyzing the difference data, this application can determine whether any components within the water pump are malfunctioning, thereby achieving early warning of potential risks to the water pump, avoiding unplanned shutdowns caused by sudden anomalies, improving the stability and reliability of water pump operation, and providing strong support for the enterprise's production process. Attached Figure Description

[0046] Figure 1 A flowchart illustrating a big data analysis and prediction method for water pump operating status provided in this application embodiment;

[0047] Figure 2 A system structure block diagram of a big data analysis and prediction system for water pump operating status provided in this application embodiment;

[0048] Figure labeling: 100, Water pump operation status big data analysis and prediction system; 101, Health parameter acquisition module; 102, Real-time parameter acquisition and difference signal generation module; 103, Data analysis module. Detailed Implementation

[0049] To better illustrate the present invention, the invention will now be described in further detail with reference to the accompanying drawings.

[0050] It should be understood that, in order to make the objectives, technical solutions, and advantages of the embodiments of this disclosure clearer, the technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. The components of the embodiments of this disclosure described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this disclosure provided in the accompanying drawings is not intended to limit the scope of the claimed disclosure, but merely represents selected embodiments of this disclosure. All other embodiments obtained by those skilled in the art based on the embodiments of this disclosure without inventive effort are within the scope of protection of this disclosure.

[0051] The following description uses at least one specific embodiment as an example. In this embodiment:

[0052] Firstly, such as Figure 1 As shown, a big data analysis and prediction method for water pump operating status is provided. This method is applied in a water pump control system and includes:

[0053] Step S1: In the application of water pump operation, obtain reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operation mode to the second water pump operation mode under healthy operation conditions.

[0054] Step S2: In the same water pump operation application, when the water pump actually switches from the first water pump operation mode to the second water pump operation mode, the real-time operation parameter data of the water pump is obtained, and the real-time operation parameter data of the water pump is matched and compared with the reference data to obtain the difference data reflecting the degree of deviation between the actual operation state of the water pump and the healthy operation state of the water pump.

[0055] Step S3: Analyze the difference data to determine if any components inside the water pump are abnormal.

[0056] By acquiring reference data on mode switching of the water pump under healthy operating conditions and comparing it with real-time data during actual operation, difference data can be generated. This enables the effective identification of abnormal states of the water pump during dynamic switching, providing a more accurate basis for predictive maintenance of the water pump.

[0057] To better understand the technical solution proposed in this application, it is necessary to explain some key terms and implementation environments involved. "First pump operating mode" and "second pump operating mode" represent the stable operating states of the pump under different working conditions. For example, both the first and second pump operating modes can be high-flow-rate or low-flow-rate delivery modes. "Operating parameters" refer to various physical quantities reflecting the pump's operating state, such as vibration, temperature, pressure, current, voltage, flow rate, and head. Real-time monitoring of these parameters is the basis for judging the pump's health status. "Reference data" refers to baseline data showing the changes of various operating parameters over time when the pump switches from one operating mode to another while in a healthy state. Examples include one or more combinations of vibration data, temperature data, or pressure data. "Real-time operating parameter data" refers to the operating parameter data collected during the same mode switching period during actual pump operation. "Difference data" is obtained by comparing real-time operating parameter data with reference data, reflecting the degree of deviation between the pump's actual operating state and its healthy operating state.

[0058] The core of the big data analysis and prediction method for water pump operation status proposed in this application lies in achieving early warning of abnormalities in internal components of the water pump through refined monitoring and comparative analysis of water pump operating parameters. Firstly, in water pump operation applications, it is necessary to obtain reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operating mode to the second water pump operating mode under healthy operating conditions. This step is crucial for establishing a health benchmark. For example, after the water pump is initially installed and commissioned and confirmed to be in optimal health, multiple tests can be conducted to switch from the first water pump operating mode to the second water pump operating mode, and the curves of various operating parameters of the water pump, such as vibration, temperature, and pressure, changing over time during each switch can be recorded. This raw data can be stored in local storage or a cloud database.

[0059] Secondly, in the same pump operation application, when the pump actually switches from the first pump operation mode to the second pump operation mode, it is necessary to acquire the pump's real-time operating parameter data. This real-time operating parameter data is then matched and compared with reference data to obtain difference data reflecting the degree of deviation between the pump's actual operating state and its healthy operating state. Specifically, when the pump control system issues a mode switching command, the data acquisition module immediately starts, acquiring various real-time operating parameter data of the pump at the same sampling frequency and duration as when acquiring the reference data. This real-time data is then transmitted to the data processing module. The data processing module performs time alignment and matching between the real-time operating parameter data and the pre-stored reference data. For example, difference data can be generated by calculating the difference between the real-time data and the reference data at each time point. This difference data can intuitively reflect the degree of deviation between the pump's current operating state and the healthy benchmark. Finally, the difference data is analyzed to determine whether any components within the pump are malfunctioning. After acquiring the difference data, the data analysis module further processes this data. For example, one or more thresholds can be set, and when one or more parameters in the difference data exceed the preset threshold, it is considered that the water pump may be abnormal.

[0060] The big data analysis and prediction method for pump operation status proposed in this application effectively identifies abnormalities in internal pump components by establishing healthy operation reference data for the pump during mode switching and comparing it with real-time data during actual operation. Traditional pump condition monitoring methods often suffer from false alarms or missed alarms due to the complexity and difficulty in accurately classifying transient data during mode switching. This application focuses on the specific dynamic process of mode switching and establishes a health benchmark for this process, making the analysis of transient data more accurate.

[0061] Compared to existing technologies, the advantage of this application lies in its ability to effectively process transient data during pump mode switching, avoiding the analytical difficulties caused by the complexity of transient data in traditional methods. By establishing health reference data for the mode switching process, this application can more accurately identify subtle changes caused by component abnormalities, thereby achieving early warning of pump abnormalities and providing a more reliable basis for preventive maintenance.

[0062] In some embodiments described above, this application proposes obtaining reference data on the changes in various operating parameters of a water pump over time when switching from a first water pump operating mode to a second water pump operating mode under healthy operating conditions. However, if only a single data acquisition is performed, the obtained reference data may be affected by accidental factors or instantaneous fluctuations, resulting in insufficient representativeness of the water pump's healthy operating state and thus affecting the accuracy of subsequent difference data analysis. Therefore, this application further proposes optimizing the method of obtaining the reference data to improve its reliability and stability.

[0063] Specifically, the step of obtaining reference data on the changes of various operating parameters of the water pump over time when switching from the first water pump operating mode to the second water pump operating mode under healthy operating conditions in the water pump operation application includes:

[0064] To obtain reference data on the changes of various operating parameters of a water pump over time when switching from the first water pump operating mode to the second water pump operating mode under a healthy operating scenario, specifically, to continuously or periodically collect various operating parameters, such as vibration data, temperature data, or pressure data, during the entire process of switching from one operating mode (first water pump operating mode) to another operating mode (second water pump operating mode) when the water pump is in a known healthy state, and record the trajectory of these parameters changing over time.

[0065] The above operation is repeated a preset number of times to obtain a dataset containing reference data showing the changes of various operating parameters of multiple water pumps over time. The preset number of times can be set according to the actual application scenario and the requirements for data stability; for example, it can be repeated 5 times, 10 times, or more. By repeatedly collecting data, multiple operating parameter curves reflecting the mode switching process of the water pump in a healthy state can be obtained.

[0066] The reference data on the changes of various operating parameters of multiple water pumps over time are accumulated and averaged to obtain the processed reference data on the changes of various operating parameters of the water pumps over time. Specifically, for each set of reference data in the dataset, the corresponding points on the time axis are accumulated, and then divided by the number of times the data is repeated to obtain an average reference data curve.

[0067] The above-described method yields more stable, accurate, and representative reference data for the healthy operation of water pumps. Compared to single-shot data collection, this method effectively reduces the impact of accidental factors and instantaneous fluctuations on the quality of the reference data, improving its resistance to interference. Consequently, when matching and comparing subsequent real-time pump operating parameter data with the reference data, the deviation between the actual operating state and the healthy operating state of the pump can be identified more accurately, providing more reliable data support for early warning and diagnosis of pump component abnormalities.

[0068] As a specific implementation method, suppose a water pump, in a healthy operating state, needs to switch from a low-speed operating mode (first pump operating mode) to a high-speed operating mode (second pump operating mode). To obtain reliable reference data, this switching process is repeated 10 times. During each switching process, operating parameters such as pump vibration, temperature, and pressure data are continuously collected and recorded. For example, during the first switching, a slight environmental disturbance might cause a slightly higher vibration value at a certain point in time, but the vibration value returns to normal during the second and third switching. By accumulating and averaging the vibration, temperature, and pressure data collected during these 10 switching processes, a smooth and stable average vibration curve, average temperature curve, and average pressure curve can be obtained. This averaged curve serves as the healthy operating reference data for the water pump switching from low-speed mode to high-speed mode.

[0069] In some embodiments described above in this application, a scheme is proposed to analyze the difference data to determine whether any components inside the water pump are malfunctioning. Specifically, the step of analyzing the difference data to determine whether any components inside the water pump are malfunctioning can be further refined as follows: fitting the difference data using a linear interpolation method to obtain a difference signal curve;

[0070] Identify the cumulative duration corresponding to curve segments that exceed a set difference threshold from the difference signal curve graph;

[0071] The cumulative duration is used to determine whether there is any abnormality in the components inside the water pump. When the cumulative duration exceeds a set duration threshold, it is determined that there is an abnormality in the components inside the water pump.

[0072] Specifically, the difference data refers to the set of data reflecting the deviation between the actual operating state and the healthy operating state of the water pump, obtained by matching and comparing the real-time operating parameter data of the water pump with reference data when the water pump actually switches from the first water pump operating mode to the second water pump operating mode. This difference data is usually discrete. To better analyze its changing trends and characteristics, a linear interpolation method can be used to fit it to obtain a continuous difference signal curve. This difference signal curve can intuitively show the deviation changes of the water pump operating parameters during the mode switching process.

[0073] Furthermore, after the difference signal curve is generated, it is necessary to identify abnormal fluctuations from it. This is achieved by setting a difference threshold; when the amplitude of a certain curve segment in the difference signal curve exceeds this threshold, it is considered that the curve segment may indicate an anomaly. After identifying these curve segments that exceed the threshold, the cumulative duration corresponding to these curve segments needs to be calculated. The cumulative duration refers to the total time that the difference signal continues to exceed the set difference threshold.

[0074] Finally, by comparing the accumulated time with a preset time threshold, it can be determined whether any components inside the water pump are malfunctioning. Specifically, when the accumulated time exceeds the preset time threshold, it indicates that the water pump's operating status deviation has been prolonged or significant, which is usually a reliable indication of an abnormality in the pump's internal components.

[0075] This application's solution, through refined processing and analysis of differential data, can more accurately identify potential anomalies in water pump operation. Traditionally, a simple threshold might be used to determine whether an instantaneous parameter is abnormal, but this method is easily affected by instantaneous fluctuations or noise, leading to false alarms or missed alarms. This solution introduces a linear interpolation method to transform discrete differential data into a continuous differential signal curve, making data trends and anomaly patterns more clearly visible. By identifying curve segments exceeding a set difference threshold and calculating their cumulative duration, short-term, occasional fluctuations can be effectively filtered out, focusing on persistent or large-amplitude anomalies. This comprehensive judgment mechanism based on duration and amplitude can more reliably reflect whether there are actual anomalies in the internal components of the water pump.

[0076] Furthermore, the current fluid state within the water pump is verified. In this embodiment, the step of determining whether any component within the water pump is malfunctioning based on the accumulated time includes, after determining that a component within the water pump is malfunctioning when the accumulated time exceeds a set time threshold, the following steps are taken:

[0077] Obtain the basic feature distribution of the differential signal curve, such as extracting its energy distribution and peak features within a specific frequency range;

[0078] If an increase in the viscosity of the currently pumped liquid is detected compared to the healthy operating state, the sensitivity parameters of the fluid in the current pump are obtained from a priori database of fluid property sensitivity parameters.

[0079] Based on the sensitivity parameters and basic characteristic distribution of the fluid in the current water pump, the reference signal curve is obtained by using the hydrodynamic equation (e.g., considering the flow resistance model of viscous fluid in the pump cavity) and combining iterative algorithm.

[0080] The difference signal curve is compared with the reference signal curve to verify whether the abnormality in the actual operation of the water flow is caused by a characteristic shift due to changes in fluid properties. For example, if the two curves are found to be highly consistent in terms of shape, amplitude, and frequency characteristics, it can be verified and confirmed that the previously identified abnormality was not caused by a malfunction in the pump components, but rather by a shift in operating parameter characteristics due to an increase in the viscosity of the transported fluid. In this way, unnecessary repairs or replacements of healthy components are avoided, thereby improving the accuracy of diagnosis.

[0081] In practical applications, abnormal pump operation can be caused not only by component wear or malfunctions, but also, distinct from the fluid state changes mentioned above, by changes in the properties of the substance being pumped, such as fluid contamination, viscosity alterations, or chemical composition variations. These changes in substance properties can also adversely affect pump performance and lifespan, but the methods described above fail to directly identify and provide early warnings for such anomalies. Therefore, this application proposes a method that, after determining that an abnormality has occurred in a component within the pump, uses entropy analysis of differential data to identify whether the properties of the substance being pumped have become abnormal, thereby providing a more comprehensive assessment of the pump's operating status.

[0082] Specifically, after analyzing the difference data to determine whether any components inside the water pump are malfunctioning, the process includes:

[0083] When it is determined that a component inside the water pump is abnormal, the entropy value of the difference data is calculated using the entropy value method, and a timestamp is assigned to the entropy value to form structured data. This structured data is then stored in the historical entropy value database.

[0084] After each water pump actually switches from the first water pump operation mode to the second water pump operation mode, the structured data stored in the historical entropy value database is extracted according to a preset time period.

[0085] Analysis of multiple extracted structural data points revealed that when the entropy value in the extracted structural data showed a continuous increasing trend over time and the slope of this continuous increasing trend exceeded a set slope threshold, it was confirmed that the properties of the material being transported in the water pump were abnormal.

[0086] Assigning a timestamp to the entropy value involves recording the corresponding time information after each entropy value is calculated, for subsequent time series analysis. Thus, the entropy value and the timestamp together form structured data, which is stored in a historical entropy database for long-term tracking and trend analysis.

[0087] In practical applications, the preset time period can be set according to the pump's operating characteristics, maintenance cycle, or empirical values, such as daily, weekly, or monthly. Each time the pump actually switches from the first pump operating mode to the second pump operating mode, the system will extract the corresponding structured data from the historical entropy database according to the preset time period.

[0088] Furthermore, analysis of multiple extracted structural data sets aims to identify trends in entropy over time. When a sustained increasing trend in entropy over time is detected in the extracted structural data, and the slope of this increasing trend exceeds a set slope threshold, it can be confirmed that the properties of the substance being transported within the pump are abnormal. The set slope threshold is used to distinguish between normal fluctuations and substantial abnormal changes, avoiding false alarms.

[0089] The above approach provides a more comprehensive capability for monitoring and predicting the operating status of water pumps. Compared to simply identifying component anomalies, this solution further enables the identification of anomalies in the properties of the substances being pumped. This allows for earlier detection and resolution of problems caused by changes in fluid properties, such as preventing product quality degradation, accelerated equipment corrosion, or increased energy consumption due to fluid contamination. Furthermore, by monitoring entropy trends, it is possible to quantitatively assess and predict changes in the properties of the pumped substances, providing data support for preventative maintenance and optimized operating strategies.

[0090] In some preferred embodiments, a specific example is given below. Suppose a water pump is used to transport a coolant, and its operating parameters under healthy operating conditions have stable reference data when switching modes. During actual operation of the water pump, due to prolonged use of the coolant or external contamination, its chemical composition and viscosity gradually change.

[0091] Each time the water pump switches from the first pump operation mode to the second pump operation mode, the system acquires real-time operating parameter data, matches and compares it with reference data, and obtains the difference data. Subsequently, the entropy value of these difference data is calculated using the entropy method, and a timestamp is assigned to form structured data, which is then stored in the historical entropy value database.

[0092] For example, in the initial stage, the coolant properties are normal, and the entropy value of the differential data remains at a low and stable level. As the coolant gradually deteriorates, its properties change, leading to increased fluctuations in the water pump operating parameters, and the entropy value of the differential data also begins to rise slowly. The system extracts structured data from the historical entropy value database according to a preset weekly time period for analysis.

[0093] When the entropy value is detected to show a continuous increasing trend for several consecutive weeks, and calculations show that the slope of this trend has exceeded a preset slope threshold (for example, the average weekly entropy value growth rate is set to exceed 0.5%), the system will immediately issue an alarm to confirm that the properties of the coolant being pumped in the water pump are abnormal.

[0094] In some of the above-described embodiments, analyzing the difference data can determine whether any components inside the water pump are malfunctioning. However, simply determining the presence of an malfunction does not directly identify the specific malfunctioning component, which hinders subsequent accurate maintenance and troubleshooting. Therefore, this application further proposes a method for identifying the malfunctioning component inside the water pump after determining that a malfunction has occurred.

[0095] Specifically, after analyzing the difference data to determine whether any components inside the water pump are malfunctioning, the process includes:

[0096] Extract prior knowledge about the relationship between pump flow rate and easily worn parts of the pump from the enterprise's pump operation database, and build a pump operation knowledge base.

[0097] When it is confirmed that the water pump is in an abnormal operating state, the water pump flow information of the current water pump in the second water pump operation mode is collected, and the associated water pump wear parts are matched from the water pump operation knowledge base based on the water pump flow information.

[0098] Analysis of easily worn parts of the water pump and differential signal curves is used to identify abnormal components within the water pump.

[0099] Specifically, in constructing a pump operation knowledge base, the wear patterns, abnormal modes, and their correlation with flow rate of various easily worn internal components (such as impellers, bearings, and seals) can be extracted and organized from the company's long-term accumulated pump operation database. This prior knowledge can include which components are more prone to wear or abnormalities under specific flow ranges or flow change trends. The pump operation knowledge base can be a structured database used to store this prior knowledge for subsequent querying and matching.

[0100] When the pump's operating status is determined to be abnormal, the system will collect the pump's flow rate information in real time under the second pump operating mode. This flow rate information can be acquired by the flow sensor in the pump control system. Subsequently, the collected pump flow rate information is matched with prior knowledge stored in the pump operation knowledge base. The matching process aims to identify the easily worn components most likely associated with the current flow rate condition from the knowledge base. For example, if the current flow rate is within a specific range, the knowledge base might indicate that the impeller or bearing is a common easily worn component at that flow rate.

[0101] In practical applications, after identifying the associated easily worn parts of the water pump, further analysis is performed using the previously generated difference signal curve. The difference signal curve reflects the degree of deviation between the actual operating parameters of the water pump and the health reference data. By combining this deviation information with prior knowledge of the easily worn parts, the range of abnormal parts can be narrowed down more precisely, thereby identifying the abnormal components within the water pump. For example, if the difference signal curve shows abnormal fluctuations at a specific frequency or time period, and this fluctuation pattern matches the typical abnormal characteristics of a easily worn part in the knowledge base, then that part can be preliminarily identified as abnormal.

[0102] This solution effectively addresses the problem in basic solutions that, while capable of identifying pump malfunctions, cannot pinpoint the specific abnormal components by introducing a pump operation knowledge base and combining it with pump flow information. Specifically, the construction of the pump operation knowledge base allows the system to accumulate and utilize historical data on the correlation between flow rate and component wear, providing a basis for preliminary location of abnormal components. When a pump is identified as malfunctioning, the collected real-time flow information serves as a key context parameter, enabling the selection of the set of easily worn components most relevant to the current operating conditions from the knowledge base. Subsequently, this narrowed set of easily worn components is comprehensively analyzed along with a difference signal curve reflecting deviations in pump operating parameters, thereby achieving accurate identification of the abnormal components.

[0103] In some preferred embodiments, a specific example is given below. Suppose that when a company's water pump switches from a first operating mode to a second operating mode, by comparing real-time operating parameter data with reference data, the difference exceeds a preset threshold, thus indicating that a component inside the pump is malfunctioning. To further identify the specific malfunctioning component, the system first extracts a large amount of historical operating data from the company's water pump operation database. This data includes abnormal patterns and wear characteristics of easily worn components such as pump impellers, bearings, and seals under different flow rates, and a water pump operation knowledge base is constructed based on this. For example, the knowledge base may record that when the water pump flow rate is 50-60 m³ / h, the probability of bearing wear is high, and its vibration signal will show abnormalities within a specific frequency range.

[0104] Once a pump malfunction is confirmed, the system immediately collects the flow rate information of the pump in the second pump operation mode, assuming the flow rate is 55 m³ / h. The system inputs this flow rate information into the pump operation knowledge base for matching, thereby identifying associated easily worn pump components. For example, the matching results indicate that bearings and seals are the components most likely to experience problems at the current flow rate.

[0105] Subsequently, the system comprehensively analyzes this information on associated easily worn components with the previously generated differential signal curves. The differential signal curves show that, in a specific high-frequency band, the vibration difference signal consistently exceeds a set difference threshold. By comparing typical abnormal characteristics of bearings and seals in the knowledge base, it was found that this high-frequency vibration anomaly pattern highly matches the early wear characteristics of the bearing, but does not match the abnormal characteristics of the seal. Therefore, the system ultimately identifies the abnormal component within the water pump as the bearing. This method allows maintenance personnel to directly inspect and replace the bearing, avoiding unnecessary disassembly and inspection of other healthy components.

[0106] Furthermore, the step of analyzing easily worn parts of the water pump and the differential signal curve to identify abnormal components within the water pump specifically includes:

[0107] The difference points exceeding the set difference amplitude are identified from the difference signal curve graph, and the corresponding water pump operating parameter data are obtained by reverse extraction based on the difference points.

[0108] The pump operating parameter data corresponding to the difference point was obtained by reverse extraction, and the abnormal components in the pump operating parameter data corresponding to the difference point were identified by time-frequency analysis.

[0109] Based on the abnormal component, the suspected wear component causing the abnormal component is obtained. The suspected wear component is matched with the easily worn components of the water pump. Based on the intersection between the suspected wear component and the easily worn components of the water pump, the component that caused the abnormality is confirmed.

[0110] Specifically, the difference signal curve is a curve obtained by matching and comparing the real-time operating parameter data of the water pump with reference data, and then fitting it using a linear interpolation method. This curve intuitively reflects the degree of deviation between the actual operating state and the healthy operating state of the water pump. The set difference amplitude refers to a pre-set threshold used to determine whether the fluctuation of the difference signal reaches an abnormal level. When the amplitude of a point or segment of the curve in the difference signal curve exceeds the set difference amplitude, it is identified as a difference point, indicating that the operating parameters of the water pump have shown significant abnormalities at that time point or time period. Based on these identified difference points, reverse tracing can be performed to accurately extract the water pump operating parameter data corresponding to these difference points from the original real-time operating parameter data, providing a raw and accurate data foundation for subsequent anomaly analysis.

[0111] Time-frequency analysis is a signal processing technique that can simultaneously analyze signal variations in the time and frequency domains, such as using short-time Fourier transform. Through time-frequency analysis, the operating parameter data obtained through reverse extraction can be decomposed into different frequency components, and the changes of these frequency components over time can be observed. Anomalies in pump components, such as bearing wear, impeller imbalance, and cavitation, typically exhibit specific frequency characteristics or transient impact signals in vibration, pressure, or temperature signals. Time-frequency analysis can effectively separate these anomaly-related components, such as specific harmonics, from complex operating parameter data, thereby revealing the type and nature of the anomaly.

[0112] In practical applications, different anomalous components often correspond to abnormal patterns in different internal pump parts. For example, specific high-frequency vibrations may indicate bearing wear, while low-frequency vibrations may be related to impeller imbalance or misalignment. By establishing a mapping relationship between anomalous components and pump component anomalies (e.g., based on expert experience, historical anomaly data, or physical models), suspected wear components that may be causing these anomalies can be inferred from the identified anomalous components. Subsequently, these suspected wear components are compared with the associated easily worn pump components matched from the pump operation knowledge base. By finding the intersection between the two lists—that is, components that appear simultaneously in both the suspected wear component list and the associated easily worn pump component list—the truly abnormal components within the pump can be identified more accurately.

[0113] This application's solution first precisely identifies the time point and amplitude of the anomaly from the differential signal curve, and then extracts the corresponding original operating parameter data, ensuring the focus of the analysis. Subsequently, time-frequency analysis is used to process this raw data in depth, effectively revealing the abnormal features hidden in the complex signals. Furthermore, by associating these abnormal features with known component anomaly patterns, suspected wear components can be preliminarily identified. Finally, cross-validation is performed using information on easily worn components related to flow rate obtained from the pump operation knowledge base, thereby confirming the abnormal component from multiple dimensions.

[0114] In some of the above-described implementations, while it is possible to identify abnormal components within the pump by analyzing discrepancy data and combining it with a pump operation knowledge base, simply identifying abnormal components may not be sufficient to support a comprehensive predictive maintenance strategy. Clear analysis and indications are not yet provided regarding the further development trends of abnormal components, their performance under different operating modes, and the evolution path of the anomaly. This may lead to delayed or inaccurate maintenance decisions.

[0115] In response, after the steps of identifying the suspected wear component based on the abnormal component, matching the suspected wear component with easily worn components of the water pump, and confirming the abnormal component based on the intersection between the suspected wear component and the easily worn components of the water pump, the process includes:

[0116] Acquire the first operating parameter data when the water pump is in the first operating mode and acquire the second operating parameter data when the water pump is in the second operating mode;

[0117] Based on the confirmed abnormal component, first abnormality information and second abnormality information about the abnormal component are extracted from the first operating parameter data and the second operating parameter data;

[0118] By associating the first anomaly information, the second anomaly information, and the difference data, a development path indication is formed for the component that has an anomaly.

[0119] Specifically, acquiring the first operating parameter data when the water pump is in the first operating mode and acquiring the second operating parameter data when the water pump is in the second operating mode refers to collecting various operating parameter data related to the water pump's operating mode when the water pump is in its preset first and second operating modes, respectively. These operating parameter data may include, but are not limited to, vibration data, temperature data, pressure data, flow rate data, current data, voltage data, etc., with the aim of capturing the performance characteristics of abnormal components under different operating conditions.

[0120] The step of extracting first and second abnormal information about the identified abnormal component from the first and second operating parameter data can be understood as follows: after identifying the specific abnormal component, information directly related to the abnormal component and reflecting its health status or abnormal characteristics is screened and extracted from the previously collected first and second operating parameter data for that component. For example, if the abnormal component is a bearing, specific data related to bearing vibration frequency and temperature changes can be extracted as abnormal information. The purpose is to focus on the specific manifestations of the abnormal component for more refined analysis.

[0121] In practical applications, the association of the first anomaly information, the second anomaly information, and the difference data to form an indication of the development path of the abnormal component specifically refers to the comprehensive analysis and correlation of anomaly information extracted from different operating modes with previously generated difference data. The difference data reflects the degree of deviation between the actual operating state and the healthy operating state of the water pump, while the first and second anomaly information provide the specific manifestation of the abnormal component under a particular operating mode. By correlating these data in multiple dimensions, such as using time series analysis or machine learning algorithms, the evolution trajectory of the abnormal component from normal to abnormal, and then to aggravated abnormality, can be constructed, thus forming a clear indication of the anomaly development path and providing a more comprehensive basis for predictive maintenance and anomaly diagnosis.

[0122] In some preferred embodiments, a specific example is given below. Suppose that during the operation of a water pump, abnormal bearing wear has been confirmed using the methods described above. To further understand the development path of this bearing wear, this application will perform the following steps:

[0123] First, acquire the first operating parameter data when the water pump is in a first operating mode (e.g., low flow rate, low speed mode), and the second operating parameter data when the water pump is in a second operating mode (e.g., high flow rate, high speed mode). This data may include vibration signals from the bearing area, temperature changes, etc.

[0124] Secondly, based on the confirmed abnormal component being the bearing, first abnormal information about the bearing is extracted from the first operating parameter data, such as the specific frequency vibration amplitude of the bearing in low-speed mode; at the same time, second abnormal information about the bearing is extracted from the second operating parameter data, such as the temperature rise trend of the bearing in high-speed mode.

[0125] Finally, these first and second anomaly reports, along with previously obtained differential data reflecting deviations in the overall pump operating status, are correlated and analyzed. For example, it can be observed that bearing-related abnormal signals persist in the differential data, and both the first and second anomaly reports show that the bearing vibration amplitude and temperature gradually increase under different operating modes. Through this correlation, a clear indication of the bearing wear development path can be formed, for example, indicating that bearing wear is accelerating and predicting that it may reach a critical abnormal state at some point in the future. Therefore, maintenance personnel can schedule repairs or replacements in advance to avoid production interruptions due to sudden bearing failure.

[0126] Secondly, such as Figure 2 As shown, a big data analysis and prediction system 100 for water pump operation status is provided, including:

[0127] The health parameter acquisition module 101 is used to acquire reference data on the changes of various operating parameters of the water pump over time when the water pump is operating healthily and switching from the first water pump operating mode to the second water pump operating mode in the water pump operation application.

[0128] The real-time parameter acquisition and difference signal generation module 102 is used to acquire real-time operating parameter data of the water pump when the water pump actually switches from the first water pump operating mode to the second water pump operating mode in the same water pump operating application. The real-time operating parameter data of the water pump is matched and compared with reference data to obtain difference data reflecting the degree of deviation between the actual operating state of the water pump and the healthy operating state of the water pump.

[0129] The data analysis module 103 is used to analyze the difference data and determine whether the operating status of the water pump is abnormal in the second water pump operation mode.

[0130] As described above, the big data analysis and prediction method for water pump operating status provided in this application acquires reference data under healthy operating conditions during water pump operating mode switching and compares it with real-time parameter data of actual operation to obtain difference data reflecting the degree of deviation between the actual operating status and the healthy operating status of the water pump. This method effectively solves the problem in existing technologies where it is difficult to accurately identify transient data in water pump rapid mode switching scenarios, leading to difficulties in data analysis and an inability to effectively predict abnormal water pump operating status. By analyzing the difference data, this application can determine whether any components within the water pump are malfunctioning, thereby achieving early warning of potential risks to the water pump, avoiding unplanned shutdowns caused by sudden anomalies, improving the stability and reliability of water pump operation, and providing strong support for the enterprise's production process.

[0131] Finally, it should be noted that the above-described embodiments are merely specific implementations of this disclosure, used to illustrate the technical solutions of this disclosure, and not to limit them. The protection scope of this disclosure is not limited thereto. Although this disclosure has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features within the scope of the technology disclosed in this disclosure. Such modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this disclosure, and should all be covered within the protection scope of this disclosure.

Claims

1. A big data analysis and prediction method for the operating status of a water pump, the method being applied in a water pump control system, characterized in that, include: In water pump operation applications, reference data is obtained showing the changes of various operating parameters of the water pump over time when switching from the first water pump operation mode to the second water pump operation mode under healthy operating conditions. In the same water pump operation application, when the water pump actually switches from the first water pump operation mode to the second water pump operation mode, real-time operating parameter data of the water pump is obtained. The real-time operating parameter data of the water pump is matched and compared with the reference data to obtain difference data reflecting the degree of deviation between the actual operating state of the water pump and the healthy operating state of the water pump. Based on the analysis of the difference data, it is determined whether there is any abnormality in the components inside the water pump.

2. The method for big data analysis and prediction of pump operating status according to claim 1, characterized in that: The reference data includes one or more combinations of vibration data, temperature data, or pressure data.

3. The method for big data analysis and prediction of water pump operating status according to claim 1, characterized in that, The step of obtaining reference data on the changes of various operating parameters of the water pump over time when switching from a first water pump operating mode to a second water pump operating mode under healthy operating conditions in the water pump operation application scenario specifically includes: obtaining reference data on the changes of various operating parameters of the water pump over time when switching from a first water pump operating mode to a second water pump operating mode under healthy operating conditions; repeating the above operation a preset number of times to obtain a data set containing multiple sets of reference data on the changes of various operating parameters of the water pump over time; accumulating and averaging the multiple sets of reference data on the changes of various operating parameters of the water pump over time in the data set to obtain the processed reference data on the changes of various operating parameters of the water pump over time.

4. The method for big data analysis and prediction of water pump operating status according to claim 1, characterized in that, The step of analyzing the difference data to determine whether there is an abnormality in any component inside the water pump specifically includes: fitting the difference data using a linear interpolation method to obtain a difference signal curve; identifying the cumulative duration corresponding to the curve segment that exceeds a set difference threshold from the difference signal curve; and determining whether there is an abnormality in any component inside the water pump based on the cumulative duration, wherein when the cumulative duration exceeds the set duration threshold, it is determined that there is an abnormality in any component inside the water pump.

5. The method for big data analysis and prediction of water pump operating status according to claim 4, characterized in that, The step of determining whether there is an abnormality in a component inside the water pump based on the cumulative duration includes, after determining that there is an abnormality in a component inside the water pump when the cumulative duration exceeds a set duration threshold, acquiring the basic feature distribution of the difference signal curve. The sensitivity parameters of the fluid in the current water pump are obtained from a prior database of fluid property sensitivity parameters; the baseline signal curve is obtained by reconstructing the fluid sensitivity parameters and basic characteristic distribution in the current water pump using the hydrodynamic equation and combined with an iterative algorithm. The difference signal curve is compared with the reference signal curve to verify whether the actual operating state of the water flow is abnormal due to the characteristic shift caused by the change in fluid properties.

6. The method for big data analysis and prediction of water pump operating status according to claim 1, characterized in that, After analyzing the difference data to determine whether any component inside the water pump is abnormal, the process includes: when it is determined that a component inside the water pump is abnormal, calculating the entropy value of the difference data using the entropy method, assigning a timestamp to the entropy value, forming structured data, and storing the structured data in a historical entropy database; after each time the water pump actually switches from the first water pump operation mode to the second water pump operation mode, extracting the structured data stored in the historical entropy database according to a preset time period; analyzing multiple extracted structured data, and when it is identified that the entropy value in the extracted structured data shows a continuous increasing trend over time and the slope of the continuous increasing trend exceeds a set slope threshold, it is confirmed that the properties of the substance being transported inside the water pump are abnormal.

7. The method for big data analysis and prediction of water pump operating status according to claim 1, characterized in that, After analyzing the difference data to determine whether any components inside the water pump are abnormal, the process includes: extracting prior knowledge about the association between the water pump flow rate and easily worn parts of the water pump from the company's water pump operation database, and constructing a water pump operation knowledge base; when it is confirmed that the water pump's operating status is abnormal, collecting the water pump flow rate information of the current water pump in the second water pump operation mode, and matching the associated easily worn parts of the water pump from the water pump operation knowledge base based on the water pump flow rate information; and analyzing the easily worn parts of the water pump and the difference signal curve to identify the abnormal components inside the water pump.

8. The method for big data analysis and prediction of water pump operating status according to claim 7, characterized in that, The step of analyzing easily worn parts of the water pump and the difference signal curve to identify abnormal parts within the water pump specifically includes: identifying difference points exceeding a set difference amplitude from the difference signal curve; performing reverse extraction based on the difference points to obtain the corresponding water pump operating parameter data; using the reverse-extracted water pump operating parameter data corresponding to the difference points to identify abnormal components in the water pump operating parameter data corresponding to the difference points using a time-frequency analysis method; identifying suspected wear parts that cause the abnormal components based on the abnormal components; matching the suspected wear parts with the easily worn parts of the water pump; and confirming the abnormal parts based on the intersection between the suspected wear parts and the easily worn parts of the water pump.

9. The method for big data analysis and prediction of water pump operating status according to claim 8, characterized in that, The steps of identifying the suspected wear component based on the abnormal component, matching the suspected wear component with easily worn components of the water pump, and confirming the abnormal component based on the intersection between the suspected wear component and the easily worn components of the water pump include: Acquire first operating parameter data when the water pump is in the first operating mode and second operating parameter data when the water pump is in the second operating mode; based on the confirmed abnormal component, extract first abnormal information and second abnormal information about the abnormal component from the first operating parameter data and the second operating parameter data; associate the first abnormal information, the second abnormal information and the difference data to form a development path indication about the abnormal component.

10. A big data analysis and prediction system for water pump operating status, characterized in that, include: The system includes a health parameter acquisition module, which acquires reference data on the changes of various operating parameters of the water pump over time when the water pump is operating healthily and switching from a first operating mode to a second operating mode. A real-time parameter acquisition and difference signal generation module is used to acquire real-time operating parameter data of the water pump when it actually switches from the first operating mode to the second operating mode in the same water pump operating application. This real-time operating parameter data is then matched and compared with the reference data to obtain difference data reflecting the degree of deviation between the actual operating state and the healthy operating state of the water pump. The data analysis module is used to analyze the difference data to determine whether there are any abnormalities in the components inside the water pump.

Images